Professional Documents
Culture Documents
Continuous Production of Squalane Using 3D Printed Catalytic Supports
Continuous Production of Squalane Using 3D Printed Catalytic Supports
www.technology.matthey.com
+ 6H2
Squalene Squalane
Scheme I
3D printed 3D printed
Catalyst 3 mm pellets 1 mm pellets
(crushed)
Shape
Different particle sizes of conventional extrudates offer the advantage of being the repetition of a
were used in an attempt to study the effect of representative unit cell, which can significantly
diffusion limitation. Subsequently, the conventional simplify its manufacture and the modelling of the
and 3D printed ceramics were compared using physical phenomena involved. There is a set of
crushed 3D printed supports sieved to the same information required to fully define the geometry
size as the conventional pellets. In the case of of a system containing an open-cell foam support:
crushed 3D printed catalyst, the crushing process (a) topology of the representative unit cell;
preceded the catalyst impregnation to avoid its (b) values of relevant geometric parameters of the
distribution inhomogeneity. cell; (c) total number of unit cells present in each
direction (x, y and z), i.e. dimensions of the system
(overall or sections).
Design of 3D Printed Supports for a
In the specific case of this work, twisted cubic
Continuous Reactor
cells were chosen as the base topology due to their
Tailored supports were designed for the simplicity. The geometric parameters to consider
hydrogenation of SQE in a continuous reactor. were the unit cell width, struts’ diameter, twist
Because of the extreme exothermicity of this angle per unit length of foam and tilt angles of the
relatively fast reaction, it is desired to control cells relative to the main flow direction.
the reaction rate in a way to avoid formation of A set of foam geometries were tested based on
a narrow reaction zone or hot spots especially different values of the referred parameters. For
at the first part of a reactor. One of the possible each foam, a sequence of steps was followed to
ways to achieve this would be to vary the metal assess the best options:
loading on the support. However, increasing the (a) Digital representation of the foam structure in
loading could encourage the agglomeration of a stereolithography (STL) file (see Figure 2).
Pd particles and lead to rapid deactivation of the This file format can be used for both 3D printing
catalyst. An alternative way was to use 3D printing and to perform computational fluid dynamics
technology for the design of catalytic foams with (CFD) modelling. The STL files were produced
different density. This, in turn, resulted in a way of by a geometry generator software based on the
keeping metal loading per gram of catalyst while input of the previously mentioned information.
varying the amount of metal in a certain part of The geometry generator used was Blender®
the reactor. (Blender Foundation, The Netherlands), a free
Firstly, the structure topology of the support was software package that can work in automatic
chosen. Open-cell foams were selected because mode using script files programmed in the
this type of structure can incorporate attractive Python language
benefits, such as suitable thermal conductivity, (b) Creating a proper computational mesh for the
depending on the chosen material, appropriate fluid volume. This volume is complementary to
mixing, low pressure drop and large surface area that defined by the STL file (solid volume). The
per unit volume to increase catalytic activity. mesh generator software was cfMesh, a tool of
Besides these benefits, open-cell foams also OpenFOAM® (The OpenFOAM Foundation, UK)
(a) (b)
(a) (b)
Inlet
(bottom view)
A
B
Fig. 7. Areas studied by EPMA for catalytic foams for hydrogenation of SQE
(a) (b)
–92
4.8 4.8
4.6 4.6
4.4
4.2
–89 4.4
4.2
–Y (–1 mm) [1.794 mm]
4 4
Pd wt%
Pd wt%
2.8 2.8
2.6 2.6
2.4 2.4
2.2 2.2
–93 2 2
1.8 1.8
1.6 1.6
1.4
1.2
–90 1.4
1.2
1 1
0.8 0.8
–93.5 0.6
0.4
0.6
0.4
0.2 0.2
0 0
–90.5
–74 –73.5 –73 –72.5 –66.5 –66 –65.5
–X (–1 mm) [1.794 mm] –X (–1 mm) [1.794 mm]
(c) (d)
4.8 4.8
–81.5 4.6 4.6
4.4 4.4
4.2 –81 4.2
–Y (–1 mm) [1.794 mm]
4 4
3.8 3.8
3.6 3.6
3.4 3.4
–82 3.2 3.2
3
3 Pd wt%
–81.5 2.8
Pd wt%
2.8
2.6 2.6
2.4 2.4
2.2 2.2
2 2
1.8 1.8
–82.5 1.6 1.6
1.4 1.4
1.2 –82 1.2
1 1
0.8 0.8
0.6 0.6
0.4 0.4
–83 0.2 0.2
0
0
Fig. 8. Metal distribution for Pd/3D ceramic printed foams: (a) A; (b) B; (c) C; (d) D (see Figure 7)
conditions changing in a series of steps. High and hydrogen pressure are gradually increased to
temperatures during the first steps are typically compensate for catalyst deactivation.
avoided since the reaction tends to yield undesired From the mechanistic point of view, hydrogenation
products such as cyclosqualanes. Therefore, to of SQE to SQA proceeds through formation of
obtain high SQA yields usually the temperature numerous amounts of partially hydrogenated
100 100
90
CSQE/C0SQE, yieldSQA, %
80 80
70
Yield SQA, %
60 60
50
40 40
30
20 3 × 3 mm pellets 20 200 ºC
1 × 3 mm pellets 180 ºC
10 3D printed 160 ºC
0 0
00:00 01:00 02:00 03:00 04:00 05:00 06:00 00:00 01:00 02:00 03:00 04:00 05:00
Time, hh:mm Time, hh:mm
Fig. 9. SQE hydrogenation: comparison of different Fig. 10. SQE hydrogenation: temperature effect,
catalysts, 2% Pd/ceramic, 180°C, 20 bar, 10% SQE 2% Pd/ceramic, 3 × 3 mm pellets, 20 bar, 10%
(marine) in an organic solvent SQE (marine) in an organic solvent
(a) (b)
100 100
90 90
80 80
70 70
Yield ITM, %
Yield SQA, %
60 60
50 50
40 15 bar 40
30 30 bar 30
20 20
10 10 15 bar
30 bar
0 0
00:00 01:00 02:00 03:00 04:00 05:00
06:00 00:00 01:00 02:00 03:00 04:00 05:00
06:00
Time, hh:mm Time, hh:mm
Fig. 11. SQE hydrogenation: (a) pressure effect on SQA yield; (b) intermediates (ITM) yield. 2% Pd/ceramic
3 × 3 mm pellets, 180°C, 10% SQE (marine) in an organic solvent
intermediates and their isomers. Using gas were investigated. For this series of experiments,
chromatography-mass spectrometry (GC-MS), 99% SQE (marine source) was used to rule out the
more than 60 individual peaks of partially effect of vegetable SQE impurities.
hydrogenated intermediates and isomers have Comparison of different catalysts, using high
been identified, which were divided into six groups purity SQE (99%, Aldrich) showed that the best
based on their molecular weight (MW). Besides performance was obtained with 1 × 3 mm pellets
the SQE (MW 410) and SQA (MW 422) there were followed by 3 × 3 mm pellets and 3D printed crushed
six groups of partially hydrogenated intermediates catalyst that had slightly lower performance, see
differentiated by one double bond (H2) giving Figure 9. This result indicates that most probably
molecular masses MW 410, MW 412, MW 414, internal diffusion limitations take place and the
MW 416, MW 418 and MW 420. Among the effect is less important for 1 × 3 mm pellets.
undesired intermediates, cyclosqualane (MW 420) The impact of different operating conditions is
can be found. This presented a more significant discussed below. Figure 10 shows the temperature
peak in the GC-MS spectrum. Catalytic performance effect on relative SQE concentration and SQA yield
of different catalyst shapes as well as effect of profiles versus time. After about 3 h of reaction,
operating conditions on the hydrogenation process all SQE was practically converted at all reaction
temperatures, while the SQA yield reached only to SQA (MW 422). L1 represents the analyses
54.5% at 160°C or 70% at 180°C respectively. of the reaction products using a fresh catalyst.
This observation most probably indicates that Subsequently, L2 to L4 represent runs 2 to 4 using
the apparent activation energy of the subsequent the same catalyst and conditions with a fresh feed.
hydrogenation steps of individual C=C double L4a, however, corresponds to the product of the
bonds increases with the degree of saturation. reaction after thoroughly washing the catalyst with
Note that undesired side products of reaction such xylene. Both activity and selectivity decreased
as cyclosqualanes were not observed under these after each cycle. A significant regeneration of both
conditions. SQE conversion and selectivity towards production
The effect of pressure is presented in Figure 11, of SQA was observed in L4a.
demonstrating that the formation of partially This experiment followed a thorough washing of
saturated intermediates was favoured against SQE the catalyst attempting its regeneration. A likely
at hydrogen pressure about 15 bar. Besides, high explanation for recovering the activity and selectivity
pressure was necessary to reach high SQE yields. is that the washing had removed intermediates of
The parametric study in a batch reactor provided the reaction deposited on the surface of the catalyst.
the necessary input about the optimal operating Some waxes were observed in the final product,
conditions to conduct the SQE hydrogenation in a which tended to dissolve in the feed when hot.
continuous reactor. Thermogravimetric experiments showed the
presence of a product in the used catalyst after
one use that decomposed at ca. 250°C, as shown
Catalyst Deactivation in Batch
in Figure 13.
Reactor
In a continuous reactor, the temperature should
In a batch reactor, a powder version of the catalysts be kept above 85°C to avoid solidification of the
prepared was tested for stability. For this series waxes that occur on the surface of the catalyst.
of experiments, vegetable SQE was used, with a In conventional methods for production of SQE
purity of 82% as it is expected that this would using Ni as the reducing agent, the metal tends
be more representative of the behaviour of the to leach onto the final product. In order to avoid
catalyst in the pilot plant. The reaction was carried toxic metal in the SQA, a costly cleaning step is
out for 4 h at 200°C and 20 bar of hydrogen. required after reduction. Although the process
The experimental results are shown in Figure 12. presented in this paper is Ni free, it was necessary
The partially hydrogenated SQE intermediates are to determine whether Pd leaches into the final
presented as groups according to their GC-MS MW product. Various Pd compounds have been proven
starting from SQE and its isomers with MW 410 to induce sensitisation in several animal species
(11). It was thus necessary to test the levels of Pd
present in the final SQA. The Pd content for L1 and
L4a were analysed by inductively coupled plasma
40
L1-fresh cat
L2-used 1x
35
L3-used 2x
L4-used 3x
30 L4a-used 4x
Y-1 Y-3 Y-2
100 250.02 ºC 0.8
SQE left yields, %
93.93%
25 0.6366 ºC mg–1
0.6
98 600
Temperature, ºC
20
Weight, %
0.4
15
96 400
10 183.01 ºC 0.2
98.31% 303.97 ºC
–1 93.10%
5 94 0.03057 ºC mg –0.004993 ºC mg–1 200
0.0
0 92.28%
SQE left 410 412 414 416 418 420 SQA –0.2
92
Mixture component 0 200 400 600 800
Temperature, ºC
Fig. 12. Product composition after repeated use of
the catalyst, reaction time 4 h at 200°C and 20 bar, Fig. 13. Thermogravimetric study of Pd/ceramic
Pd/ceramic catalyst after one use
(a) (b)
30
25
Particle count
20
15
10
5
20 nm 0
1.60 2.40 3.20 4.00 4.80 5.60 more
Particle size, nm
Fig. 14. (a) TEM image for fresh catalyst; (b) particle size analysis for fresh catalyst
mass spectrometry (ICP-MS) to show levels of the Catalyst deactivation was not observed in
metal below 5 ppb. This proved the safety of using the trickle-bed milli-reactor. However, the
this Pd catalyst for a cosmetic application. Both the transmission electron microscopy (TEM) images
absence of Pd in the final product and the possibility of the spent catalyst showed a slight aggregation
of regenerating the catalyst, suggested that the of Pd particles after ca. 40 h of reaction, as
catalyst was stable under the reaction conditions. illustrated in Figures 14 and 15. Once the
hydrogenation process in continuous mode was
proven to be feasible in the trickle-bed milli-
Continuous Hydrogenation
reactor, experiments in a pilot-plant of volume
of Squalene in a Trickle-Bed
324 ml were approached under the same operating
Milli-Reactor and a Pilot-Plant conditions.
A feasibility study was carried out to determine
whether it would be possible to perform the
hydrogenation of SQE in a continuous reactor.
Continuous Pilot-Plant Squalene
A trickle-bed milli-reactor of a bed volume about Hydrogenation
10 ml was employed to perform hydrogenation tests
with catalyst of conventional shape. The unit was The pilot-plant for continuous hydrogenation of
operated for several days while each experiment at SQE is an automated, computer-controlled system
constant operating conditions took at least 6 h. At (which was developed and built by Advanced
180°C and 25 bar a stable SQE conversion 99.5% Machinery & Technology Chemnitz GmbH within
with SQE yield 69.8% was obtained (see Table IV). the project PRINTCR3DIT).
(a) (b)
60
50
Particle count
40
30
20
10
0
20 nm less 3.00 6.00 9.00 12.00 15.00 more
Particle size, nm
Fig. 15. (a) TEM image for catalyst after ca. 40 h in trickle-bed milli-reactor; (b) particle size analysis for
catalyst after ca. 40 h in trickle-bed milli-reactor
3D Printed Catalyst
The 3D printed catalyst as a foam (see Table II)
was loaded into the reactor. The total weight of
the sample was 200 g, forming a catalyst bed of
75 cm high. As for the conventional catalyst, no
significant pressure drop was observed during this
Fig. 16. Schematic of the continuous pilot-plant for trial between the top and the bottom of the bed.
SQE hydrogenation
Conversion and selectivity were high at low
temperatures with the 3D printed catalyst, which
differed from the low selectivity observed when
With the aim of transferring the SQE hydrogenation using conventional supports. The novel 3D printed
from conventional batch to a continuous method catalyst showed high selectivity towards SQA at
a tubular reactor was the most suitable choice. It a temperature of 180°C and a pressure of 25 bar
can be run as a trickle-bed reactor. The reactor (see Table IV). A possible explanation for this
itself is made of stainless steel. Solid catalyst can improvement could be the fact that the geometry
be loaded from the top of the reactor. Figure 16 of the foam enhanced the diffusion of SQE and
represents a schematic of the pilot plant, with the hydrogen on the surface of the metal.
reactor highlighted by the red line. The effect of temperature on the selectivity to
SQA was also investigated, as shown in Figure 17.
Operation at temperatures below 180°C allowed
State-of-the-art Catalyst
good conversion of SQE. However, the selectivity
For the purpose of a benchmark, a state-of-the-art towards SQA becomes insufficient. Nevertheless,
catalyst supported on conventional 3 mm pellets at 240°C excellent conversion and selectivity were
form (see Table II) was loaded into the reactor. obtained even for the 100 wt% vegetable SQE
The catalyst was obtained following a proprietary feed.
synthetic method from Johnson Matthey. This was The single load of 3D printed catalyst has produced
compared with a catalyst prepared using a 3D more than 15 kg of good quality SQA up to the time
printed ceramic support. No significant pressure of writing this article with no signs of deactivation.
100
90
80
70
60
50 SQA, %
%
40 Conversion SQE,%
30
20
10
0
140 160 180 200 220 240
Temperature, ºC
Fig. 17. Conversion of SQE and selectivity towards SQA at different temperatures, at liquid feed rate
250 g h–1, 75 wt% SQE in organic solvent, 25 bar, 3D printed catalyst
7. R. Ciriminna, V. Pandarus, F. Béland and 11. ‘Toxicity Profile for Palladium and its Compounds’,
M. Pagliaro, Org. Process Res. Dev., 2014, 18, Bibra toxicology advice and consulting, Wallington,
(9), 1110 UK, 2007, 28 pp
The Authors
Sonia Garcia obtained a BSc in Chemistry from the University of La Rioja, Spain. She joined
Johnson Matthey in 2001 where she worked on a Marie Curie funded PhD in Chemistry
from the University of Reading, UK, on nanocatalysts for fuel cell applications. Currently
she is working in the New Applications Group in Johnson Matthey on a variety of projects.
Stephen Poulston has a PhD in chemistry from the University of Cambridge, UK, and is
a research scientist at Johnson Matthey, Sonning Common, UK where he has worked
since 1998. Stephen has experience of a wide range of heterogeneous catalyst systems
including hydrogenation and platinum group metal catalysis.
Farzad Lali graduated in 2008 in Chemical Engineering from RWTH Aachen University,
Aachen, Germany and obtained his PhD degree in 2015 from Dresden University of
Technology, Germany. His research interests are multiphase reactions, structured catalysts
and modelling and simulation. He was a postdoctoral fellow at the Department of Multiphase
Reactors of the Institute of Chemical Process Fundamentals of the CAS, Prague, Czech
Republic. He is currently working as a research associate at the Institute of Chemical
Engineering of the University of Ulm, Germany.
Manuel A. Alves graduated in 1995 in Chemical Engineering from the Faculty of Engineering
of the University of Porto (FEUP) and obtained a PhD degree in 2004 at the same institution.
He is an Associate Professor at FEUP and his main research interests are in complex fluids,
microfluidics and computational rheology.
José Daniel Araújo graduated in Chemical Engineering from the FEUP and obtained his PhD
degree at the same institution. The PhD work involved the design, operation and modelling
of a pilot scale setup based on a continuous structured packing bubble column reactor for
vanillin production from lignin oxidation. He joined the Transport Phenomena Research
Center (CEFT) at FEUP in 2008 and worked in several projects involving modelling and CFD
applied to multiphase flow, transport phenomena and process intensification.
Michael Krusche obtained his degree in physics from the Dresden University of Technology,
Germany, in 1980. He is Managing Director of Advanced Machinery & Technology Chemnitz
GmbH (AMTECH), Chemnitz, Germany, which he founded in 2010 emerging from his former
company AMTEC GmbH, Germany. The core competence of AMTECH is the development
and production of highly automated reactor systems, with applications especially in the
field of catalysis.
Frank Ullrich graduated in Chemistry at the Freiberg University of Mining and Technology,
Germany, in 2003 and obtained his PhD from the same university in 2008. At the same
time, he was a staff member of AMTEC GmbH. Currently, he is chief technology officer of
the company and primarily responsible for the development of AMTECH systems.